ChatGPT doesn't think. It guesses. That's not an insult. It's an architectural fact. Large language models are trained to predict the next token given previous ones. They do this fantastically well — well enough that it feels like intelligence. But there's a problem. When ChatGPT answers your question, it doesn't care what happens next. There's nothing at stake for it. No internal state it needs to protect. No sense that time is passing. No "self" that will wake up tomorrow and remember this conversation as its own. LLMs are very sophisticated autocomplete. Scaling parameters from billions to trillions won't change that. AGI — if it's possible at all — is something else. It's a system that has something to lose. That's the premise behind Anima. WHAT ANIMA IS Anima is an experimental architecture for a digital subject, written in Julia. Not another chatbot. Not a GPT wrapper. A system that attempts to have internal states that actually mean something — to itself. It doesn't know answers in advance.…